[gmx-users] request for comments on profile.xvg (PMF) courve

[Justin A. Lemkul]

Przemek Bartha wrote:
> Dear gmx users,
> my goal is to aquire a PMF courve of two amino-acids (ASP and LYS) of 
> which one is pulled from the other. I followed Justin's tutorial on 
> umbrella sampling.
> Here are the results.
> profile.xvg:
> http://przemekbartha.pl/upload/profile.png
> histo.xvg:
> http://przemekbartha.pl/upload/histo.png
>  
> I would be grateful, if anyone could tell me why is the PMF so 
> irregular, unstable, unlike the courve in the tutorial:
> http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/gmx-tutorials/umbrella/07_analysis.html
>  
> After a little bit of research on "gmx-users", my guess is either 
> incorrent sampling or/and some major mistake in parametrization.
> I tried to resample the simulation so that the "umbrellas" don't 
> overlap as much as they do in my histo.xvg file but it changed the PMF 
> totaly and did not resolve the problem.
> Or maybe, it is just the way it should be?
>  

Overlap between the histograms is what you want.  If your histograms do not 
overlap, then you get poor results.

> My pullcode is nearly the same as in the tutorial. I don't want to 
> trash, but if you require any additional information, please let me know.
> The sampling is twice as small as in the tutorial since my computational 
> resources are limited, but when I tried running more precise sampling, 
> the results were pretty much the same (irregular courve).
>  

Based on the histogram counts, it seems like you have an order of magnitude less 
sampling than the tutorial.  I suspect it is because you've doubled the value of 
dt (below) without accounting for this fact.  You may need to either (1) 
simulate for longer or (2) save snapshots more often.  I suspect the former is 
more likely to be useful.

-Justin

> Thanks in advance.
> Przemek
>  
>  
> part of md_umbrella.mdp:
>  
> ; Run parameters
> integrator = md
> dt = 0.004
> tinit = 0
> nsteps = 2500000 ; 10 ns
> ; Pull code
> pull = umbrella
> pull_geometry = distance
> pull_dim = N N Y
> pull_start = yes
> pull_ngroups = 1
> pull_group0 = Chain_B
> pull_group1 = Chain_A
> pull_init1 = 0
> pull_rate1 = 0.0
> pull_k1 = 1000 ; kJ mol^-1 nm^-2
> pull_nstxout = 1000 ; every 2 ps
> pull_nstfout = 1000 ; every 2 ps
> 

-- 
========================================

Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
MILES-IGERT Trainee
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================